Some meteorological and physiographic aspects of winter precipitation variation in... southwestern Montana

advertisement
Some meteorological and physiographic aspects of winter precipitation variation in mountainous
southwestern Montana
by Larry Eugene Holman
A thesis submitted to the Graduate Faculty in partial fulfillment of the requirements for the degree of
MASTER OF SCIENCE in Earth Science
Montana State University
© Copyright by Larry Eugene Holman (1971)
Abstract:
Precipitation data from a mesoscale network have been collected in conjunction with a weather
modification program in the mountains of southwestern Montana during the winter of 1969-70.
Stepwise multiple regression techniques were used to. examine the relationship between
meteorological and physiographic parameters and precipitation amount, intensity, and frequency.
Meteorological data were extracted from rawinsonde and radiosonde records. The investigation of
relationships between meteorological factors and precipitation received primary emphasis.
Stratification of precipitation data according to temporal and spatial parameters was used to increase
the multiple correlation coefficients in the meteorological analysis. Several factors appeared significant
in the regression analysis. However, cloud base temperature and 500 millibar temperature were
selected as the most important factors related to precipitation because of their frequent and often most
significant contributions to the regression equations.
In the physiographic analysis, data was extracted from USGS topographic maps and canopy
photographs of gage sites. The analysis showed that elevation exhibited a definite control on
precipitation intensity and frequency. STATEMENT OF PERMISSION TO COPY
In presenting this thesis in partial fulfillment of the
requirements for an advanced degree at Montana State University, I
agree that the Library, shall make it freely availible for inspection.
I further agree that permission for extensive copying of this thesis
for scholorly purposes may be granted by my major professor, or, in
his absence, by the Director of Libraries.
It is understood that any
copying or publication of this thesis for financial gain shall not be
allowed without my written permission.
S OM E M E T E O R O L O G I C A L AND P H Y S I O G R A P H I C A S P E C T S OF W I N T E R P R E C I P I T A T I O N
VARIATION IN MOUNTAINOUS SOUTHWESTERN MONTANA
- _
by
LARRY EUGENE HOLMAN
A thesis submitted to the Graduate Faculty in partial
fulfillment of the requirements for the degree
of
MASTER OF SCIENCE
in
Earth Science
Approved:
Head
Major 6 ejpartment
Chairman, Examining Committee
• ■ MONTANA STATE UNIVERSTlY
Bozeman, Montana
Dune, 1971
ACKNOWLEDGEMENTS
Appreciation is extended to the Bureau of Reclamation for
providing the grant which made this study possible and to the
Department of Earth Sciences, Montana State University, for making
a portion of that grant available for this research.
Dr. Val Mitchell is accorded special appreciation for his
interest, supervision, and editorial help while serving as chairman
of this writer’s Graduate Examining Committee.
Appreciation is also
extended to Dr. Arlin Super and Mr. Robert Yaw 9 who both critically
reviewed and corrected the manuscript, and to Mr. Milton Edie for
his aid and interest.
Special gratitude is extended to my wife, Eileen, and to my
family for their encouragement, patience, and understanding during
the preparation of this thesis.
The writer is also indebted to
Connie Shafer for,her typing of the final manuscript.
TABLE OF CONTENTS
Page
VITA . „ ........
ACKNOWLEDGEMENT
............ ii
.
......... ill
TABLE OF CONTENTS
............. iv
LIST OF TABLES . .
o •• o »
vi
LIST OF FIGURES
• «o e «
Vil
.
ABSTRACT ........
, e. . e viii
Chapter
3.
....................
A.
Nature and Purpose of Study . . . . .
B.
Review of Literature
I
1-
............
4
COLLECTION . . . . . . . . . . . . . . . . . . . . .
11
A.
Precipitation Compilation........ .............
11
B.
Compilation of Meteorological Factors
12
C.
Compilation of Physiographic Data. . . . . . . . .
DATA
. . . . .
....
. . . . .
........
. . . . . . .
DATA
14
ANALYSIS
22
A.
General Remarks . . . . . . . . . . . . . . . . . .
22
B.
Meteorological Aspects
24
Ce
D.
4.
. . . . . . .
.
2.
INTRODUCTION
................
. . . . .
Physiographic Aspects . . . . . . . . .
.
1.
Precipitation Frequency with Time and Space . . . .
CONCLUSIONS AND RECOMMENDATIONS.............
45
38
40
V
Chapter
Page
A»
Conclusions * * * * * # » # # * # * @ @ @ * 0 * * *
45
B«
Recommendations............. ..
46
DETERMINATION OF CLOUD DATA FROM RADIOSONDE
OBSERVATIONS ....................................
50
APPENDIX As
LITERATURE CITED . . . . . .
........
. ......................
61
LIST OF TABLES
Table
Title
Page
2,1
Radiosonde Data Summary . . . . . . . . . . .
2.2
Rawinsonde Data Summary . . . . . . . . . . . . . . . . .
16
2.3
Summary of Physiographic Factors Associated with Gage
Locations .......... . . . . . . . . . . . . . . . . . .
20
Maximum Surface Pressure Gradient Direction
Stratification . . . . . . . . . . . . . . . . . . . . .
26
3.1
........
.
15
3.2
Elevation Stratification Groups . . . . . . . .
3.3
Elevation Stratifications . . . . . . . . . . . . . . . .
29
3.4
Precipitation Cases for Each Stratification . . . . . . .
31
3.5
Important Meteorological Factors Selected by Stepwise
Multiple Regression . . . . . . . . . . . . . . . . . . .
33
.........
3.6
Physiographic Influences on Precipitation ...............
3.7
Variation of Precipitation Frequency
. . . . . . . . . .
28
39
42
LIST OF FIGURES
Figure
Title
£122
1.1
Experimental Area and Gage Network . . . .
1 .2
East-West Topographic Cross Section of
Experimental Area . . . . ........ . . . . . . « . ♦
..........
.
3
.
5
51
1.3
Season Precipitation Totals at Bozeman 12NE
for the Period 1950-70 . ........ . . . . . . . . . .
1.4
Season Precipitation Totals at Livingston
for the Period 1950-70 . . . . . . . . . ............
A.1
Cloud Boundaries Determined by Relative Humidity . . . .
A.2
Relation Between Frost Point and Dew Point
Mixing Ratios ...................... . . . . . . . .
A.3
Cloud Boundaries on Original Radiosonde Humidity Trace .
54
A.4
Minimum Relative Humidity for Cloud Determination
57
A.5
Unsteady Relative Humidity Within a Cloud
, . .
. . . . . .
.
60
ABSTRACT
Precipitation data from a mesoscale network have been collected
in conjunction with a weather modification program in the mountains of
southwestern Montana during the winter of 1969-70. Stepwise multiple
regression techniques were used to. examine the relationship between
meteorological and physiographic parameters and precipitation amount,
intensity, and frequency.
Meteorological data were extracted from rawinsonde and radio­
sonde records. The investigation of relationships between meteorolog­
ical factors and precipitation received primary emphasis. Stratifica­
tion of precipitation data according to temporal and spatial parameters
was used to increase the multiple correlation coefficients in the
meteorological analysis. Several factors appeared significant in the
regression analysis. However, cloud base temperature and 500 millibar
temperature were selected as the most important factors related to
precipitation because of their frequent and often most significant
contributions to the regression equations.
In the physiographic analysis, data was extracted from USGS
topographic maps and canopy photographs of gage sites. The analysis
showed that elevation exhibited .a definite control on precipitation
intensity and frequency.
I.
INTRODUCTION
Ac
Nature and Purpose of Study
In mountainous regions like the western United States, spring
and summer streamflow is largely dependent upon the snowpack accumu­
lated at high elevations.
Due to increasing water needs, studies on
precipitation augmentation are being undertaken in several sections
of the United States.
One such study is being conducted by Montana
State University in the Bridger-Bangtail region of southwestern
Montana (Mitchell,•jet jal., 1970).
As a part of the precipitation
augmentation program, there is a network of 21 weighing-type precip­
itation gages covering an estimated 300 square mile experimental area.
With an additional six gages generally downwind of the experimental
area, a total of 27 stations comprise the mesoscale network which
provided the precipitation data for this study.
The needs of a precipitation augmentation study justify a
concurring investigation of the areal and temporal variations in
precipitation to provide additional information for analyzing the
results of the seeding program.
Clarifying the role of various
meteorological and physiographic parameters in wintertime precip­
itation also compliments the efforts of a precipitation augmenta­
tion study.
Therefore, isolating various meteorological and
physiographic parameters, and relating their individual or combined
-2
sffects on precipitation amounts is the problem under consideration®
As related to mountainous southwestern Montana during the
winter, the specific objectives of this study are:
1«
To determine the feasibility of using stepwise multiple
regression techniques to analyze precipitation,
2,
To determine the effect of meteorological parameters on
precipitation amounts,
3,
To determine the effect of physiographic parameters on
precipitation amounts,
4,
To determine the variation in precipitation frequency and
intensity with time and space within the limits of the
study area,
As noted earlier, the mesoscale network includes 27 precip­
itation gages.
Twenty-one of the gages are concentrated in the 300
square mile experimental area.
The additional six gages provide data
east (generally downwind) of the experimental area.
The experimental
area, approximately 20 miles in length and TB miles wide, is oriented
in an east-west direction.
The western half of the area consists of
the Bangtail Ridge portion of the Bridget Mountain Range,
This
mountainous terrain grades off into the Shields Valley to the east.
West of the Bangtail Ridge lies the north-south oriented main ridge
of the Bridget Range,
Figure 1,1 is a contour map of the area showing
the approximate locations of the precipitation gages.
An east-west
EXPERIMENTAL
AREA
AND GAGE NETWORK
UVlNQSTOl
Network Gage
Season Isohyets
Elevation Contours -—
Galloiin Mis.
Figure 1.1
Isohyets represent 1969—70 seasonal totals for the network shown
in inches of water equivalent.
Isohyets are
-4'
cross section of the area is given in Figure 1.2,
The target area is characteristically dominated by westerly
flow during the winter.
However, northeasterly flow in the lower
layers of the atmosphere is a frequent occurrence.
Precipitation
from the westerly storms increases noticeably with elevation whereas
precipitation resulting from the northeasterly flow increases less
strongly with elevation.
The ratio of northeasterly to westerly
mean per storm precipitation typically is about 4:1 in the valley and
2:1 in the mountains (Super, et al.. 1968).
Such ratios suggest the
northeasterly flow is more important to valley stations than westerly
flow.
As a regional approximation for average precipitation during
the November through March season, Figures 1.3 and 1.4 show the
seasonal totals for the past twenty years from two USWB stations near
the target area.
The locations of the respective USWB stations
supplying the data are shown in Figure 1.1.
Also shown in Figure 1.1
are the isohyets of 1969-70 network precipitation totals,
B.
Review of Literature
Variations in precipitation amount, intensity, and duration
have been studied by several authors.
It is realized that not every
study has the same design and, therefore, the individual studies
cannot be objectively compared with each other.
Consequently attempts
to relate precipitation amounts with meteorological factors have
ELEVATION
(Thousands of Feet
msl)
EAST-W EST TOPOGRAPHIC CROSS
OF EXPERIMENTAL AREA
BRIDGER
MAIN
RIDGE
MOUNTAINS
BANGTAIL
RIDGE
SHIELDS
DISTANCE
Figure 1.2
SECTION
(Miles)
VALLEY
SEASON PRECIPITATION TOTALS AT
BOZEMAN I2NE FOR PERIOD 1950-70
<
O
f—
N
CL X
O
O Z
2 0 YEAR
AVERAGE
LU
CZ
CL
I
Z
O
CO
<
IU
CO
Figure
CTi
I
20-1
Figure 1.4
SEASON PRECIPITATION TOTALS AT
LIVINGSTON FOR PERIOD 1950-70
™Qw
varied in success and conclusion.
However, some insight into mountain
precipitation processes can be gained by considering those studies
which have been done.
Meteorological factors as defined in this study are factors
measured by radiosonde or rawinsonde observations,
Chappell (1967)
and Mooney and Lunn (1969) found that upper air parameters were
important in determining the effectiveness of cloud seeding.
Since
such meteorological factors are significant in precipitation augmen­
tation, interest concerning their effect on natural precipitation
processes is stimulated.
Yaw, et al, (1969) objectively tabulated winter precipitation
for southwestern Montana by 700 millibar wind direction and maximum
surface pressure gradient direction.
Three groupings of precipitation
events were noted, each representative of a particular synoptic situ­
ation,
The three groups were low pressure systems, westerly flow, and
northeasterly flow.
Each group suggested precipitation was partially
dependent upon direction of surfacial flow.
In another synoptic scale study, Klein (1967) emphasized the
importance of the closed low aloft over the western United States,
Precipitation amounts tend to increase in and just east of closed lows
aloft,
Klein suggests that factors such as fronts, vorticity, advec-
tion, thickness, and moisture also need to be considered in precip­
itation prediction studies.
■9
The results of these studies cite the importance of upper a i r ,
parameters in relation to precipitation.
However, the basic precip­
itation physics (convection, nucleation, etc,) are not to be ignored.
Physiographic factors, those agents resulting from terrain
features, cause variations in precipitation amounts and frequencies.
Previous research in the Rocky Mountain region has shown that precip­
itation generally increases with elevation (Packer, 1962),
The
literature has also indicated that other physiographic influences
contribute to the variation in gage catch in mountainous areas
(Spreen, 1947),
Among these were slope, aspect, and orientation.
Brown and Peck (1962) developed a subjective classification
system to evaluate the reliability of measurement of precipitation
occurring primarily as snow in relation to gage exposure.
Their
classification noted differences in precipitation amounts due to
vegetation and terrain features near the gage.
They also suggested
further development of an objective technique for classifying precip­
itation gage sites*
This study, along with the others cited, rein­
forces the supposition that physiographic factors are important
considerations for explaining precipitation variation.
Precipitation events as related to precipitation totals have
also been studied,
As used here, a precipitation event is the
recording of 0,0 1 or more inches of precipitation at a particular gage
during a 24 hour period,
Marlatt and Riehl (1963) noted that half of
-IDthe annual precipitation total in the upper Colorado River Basin was
realized during 16 per cent of the precipitation events.
They also
found a high correlation between the number of precipitation events
greater than 0,10 inch and the annual totals,
As a result of their
study they suggested that any precipitation less than 0 ,1 0 inch could
be considered "noise,"
In Vermont, Engman and Hershfield (196B) found increases in
the frequency of precipitation events with elevation.
This implies
that the increase in precipitation totals with elevation is due, at
least in part, to more precipitation events at higher elevations.
The studies cited above point out that not only does precip­
itation in mountainous regions vary greatly in time and space, but
that many factors contribute to that variation.
The purpose of this
study is to attempt to gain a greater understanding of the role of
several factors in precipitation processes and variability in a
mountainous region.
II
Ac
D ATA C O L L E C T I O N
Precipitation Compilation
Precipitation data was summed over six hour intervals to
correspond with radiosonde launches,.
In this way a radiosonde record
would represent conditions for a corresponding precipitation record.
Only periods having a record of precipitation were considered in the
analysis.
A record of precipitation was considered to be a six hour
period registering 0 .0 1 or more inches of precipitation at one or more
of the 27 network gages.
There were three cases when only one station
recorded precipitation during a six hour interval and there were six.
cases when all 27 stations recorded precipitation.
But regardless of-
the number of stations which registered precipitation during a six
hour period, that period was considered to be a precipitation record
and was used in the analysis.
The non-seeded cases were not separated from the seeded cases
since the resulting sample size would be too small for valid statis­
tical analysis.
The possible effect of seeding was one of many
variables not considered in the investigation.
All precipitation
totals were extracted from Weighing-type precipitation gage records.
The weighing-type precipitation gage used in the network is a
recording instrument whose function is to provide a record of precip­
itation intensity, time of occurrence, and precipitation totals.
All
gages were a modified USliiB type with 11 5/16 inch diameter orifices
a
■
2”
and one rotation per day clocks yielding time and water equivalent
resolutions of +5 minutes and ±0.005 inches respectively (Super, et
jal,, 1968),
Gages were equipped with modified Alter shields.
Precip­
itation amounts were totaled for six hour intervals over a 150 day
season from 1200MST, November 1, 1969, to 1200MST, March 31, 1970.
Also compiled for each station were records of precipitation events
and seasonal totals.
It should be noted that at times interpolation
techniques were employed to make estimates for missing six hour totals.
Less than two per cent of the precipitation data were missing.
Estimates were made subjectively, using precipitation amounts from
nearby gages as a guide.
Three basic data summations resulted from this compilation.
The first consisted of six hour precipitation amounts.
From this
primary assemblage, seasonal totals and precipitation events were
recorded.
These final collections provided the dependent variables
for statistical analysis,
B,
Compilation of Meteorological Factors
There were two major sources of meteorological data, the first
being rawinsonde records from Great Falls, Montana (GTF), approximately
100 miles north of the study area.
Secondly, as part of the weather
modification activities, a local radiosonde program helped provide
data.
Radiosonde observations were made in accordance with the
operational activities of the seeding program.
Launches were made
— 13 “
only if prescribed weather criteria were met.
The USWB station at
Helena, Montana, provided a special precipitation probability forecast
for Bozeman, Montana.
If the precipitation probability exceeded
20 per cent, the cloud cover over the target area was greater than
5/10 (broken or overcast), and the cloud base over the target area was
less than 1 0 ,0 0 0 feet msl, then an operational period commenced
(Super, et al., 1969).
Radiosonde launches were made at six hour
intervals during an operational period, if the weather criteria
persisted.
The GTFi rawinsonde data collected were height, temperature,
dew point temperature, wind direction, and wind speed for the 400,
500, and 700 millibar levels.
In some cases temperature, wind direc­
tion, dew point temperature, and wind speed values were missing.
Interpolation techniques were employed to determine missing tempera­
ture, dew point temperature, and wind direction.
scale was used to estimate wind speed.
A geostrophic wind
Because many 400 millibar dew
point temperature values were missing, this parameter was eliminated
from the statistical analyses.
The 400 mb, 500 mb, and 700 mb data
(heights, temperatures, and dew point temperatures) from the local
radiosonde records were not used because of the inconvenience and time
needed, for extracting the information.
/
Since precipitation amounts were totaled over six hour
intervals, one GTF rawinsonde record was assumed to represent
*14conditions for a twelve hour period or the equivalent of two six hour
precipitation intervals.
The OSOOMST and 1700MST launches provided
data for the periods 0000M5T to 1200MST and 1200MST to 2400MST
respectively.
The soundness of this assumption is open to question,
but more frequent rawinsonde data were not available.
The information taken from the local radiosonde soundings were
cloud base and cloud top heights and the respective temperatures and
dew point temperatures.
Cloud thickness, cloud lapse rate, and
temperature difference between cloud base and cloud top were then
derived.
An objective method was developed to determine the location of
cloud base height and cloud top height.
is given in Appendix A.
analyzed.
An explanation of this method
A total of 150 radiosonde records were
Thirty-four of the records were discarded because no clouds
were indicated.
Of those analyzed, Table 2.1 lists the mean, median,
and standard deviation for each parameter extracted from the radio­
sonde records.
Table 2.2 lists the corresponding values for the
rawinsonde parameters.
C.
Compilation of Physiographic Data
The term physiographic factor, as used here, means a topographic
or vegetation feature that may influence precipitation variation.
The
physiographic factors used in this study are defined below.
Station Elevation - the height in feet msl of the gage location.
T A B L E 2.1
R A D I O S O N D E DAT A S U M M A R Y
Factors
Mean
Median
.Standard
Deviation
Cloud top height (millibars)
602.6
594.0
63.0
Cloud top temperature (°C)
-2 0 .0
-19.9
6 .8
Cloud base height (millibars)
783.9
800.0
67.6
Cloud base temperature (°C)
-9.1
-9.2
5.9
Cloud thickness (millibars)
181.3
165.5
89.2
10.9
10.7
Lapse rate within cloud (°C/millibar)
.0635
.0699
.0396
600 millibar temperature (°C)
-18.8
-17.3
5.2
Temperature difference within cloud (0 C)
.
6.9
T A B L E 2.2
RAWINSONDE DATA SUMMARY
Factors
Mean
Median
Standard
Deviation
700 mb height (feet msl)
9666.2
9681.0
2 1 0 .2
700 mb temperature (0 C)
. - 1 2 .2
-11.4
8.4
700 mb dew point temperature (0 C)
-16.5
-15.2
10.7
700 mb wind direction (degrees)*
172.3
16G1.0
85.1
24.7
25.0
11.2
17,875.1
17,880.4
321.8
500 mb temperature (°C)
-27.4
-26.9
5.5
500 mb dew point temperature (0 C)
-33.7
-33.2
7.8
500 mb wind direction (degrees)*
163.6
160.0
67.2
41.8
40.0
16.2
22,811.7
23,014.8
229.9
400 mb temperature (°C)
-38.6
-38.0
6 .6
400 mb wind direction (degrees)*
163.5
160.0
74.5
63.0
57.0
78.2
700 mb wind speed (knots)
500 mb height (feet msl)
500 mb wind speed (knots)
400 mb height (feet msl)
400 mb wind speed (knots)
*The units of all wind direction values are degrees counterclockwise from
southeast. This transformation was performed on all wind directions to create
normal distributions.
Half Mile Elevation - the average elevation in feet msl for the
area within one-half mile of the station.
A circular dotted grid with
radius equivalent to one-half mile was overlain on USGS topographic
maps with scales of 1:24,000 and 1:62,500.
By extracting elevation ■
values for each dot on the grid, an average elevation for the area was
calculated.
The grid had 35 dots regularly spaced.
Four Mile Elevation - the average elevation in feet msl for the
area within four miles of the station.
The same procedure as
discussed for half mile elevation was utilized to compute four mile
elevation.
Directional Elevation - the average elevation of a quadrant in
each of four cardinal directions from the station figured at one-half
and four miles.
The units on this parameter were feet msl.
The
directional elevation was evaluated in a manner similar to that of
average elevation with the dotted grid divided into direction quad­
rants.
In this way upslope and downslope characteristics with respect
to the gage location would be represented.
Contour Orientation - the number of intersections resulting
between the contours of a topographic map and an overlying grid of
equally spaced parallel lines.
The series of lines was oriented on
a topographic map in each of four directions (north-south, northeastsouthwest, east-west, or northwest-southeast).
Each intersection made
by the crossing of a contour by a grid line was counted.
The sum of
-
18 -
ths intersections represented the contour orientation value.
This
parameter was computed at the one-half and four mile scales for each
direction listed.
Roughness - the sum of the four contour orientation values.
This
parameter was computed at both the one-half and four mile scales.
Vertical Canopy - measurement of the canopy less than 45° from
vertical over each gage.
A vertical photograph was taken with a pin
hole camera from the orifice of each gage.
from the photograph,
the photograph.
Percent canopy was computed
A grid of regularly spaced dots was placed on
By counting the number of dots coincident with canopy,
a percentage figure was calculated*
Horizontal Canopy - "A sequence of eight photographs, the first
oriented North and the remaining proceeding clockwise at 45° incre­
ments, was taken at each site using a tripod mounted 35 mm camera
equipped with a wide angle lens.
Sufficient overlap was obtained to
trim and assemble the enlarged black and white prints (3 X 4.5 inches)
into a strip showing the 360 degree horizontal view of terrain and
vegetative cover around the site.
Calibration photographs established
that the subtended angle between the center and the top margin of the
prints was 20 degrees" (McPartland, 1971).
■ Percentage of area in each quadrant from the horizontal to 20
degrees above, blocked by terrain or vegetation cover, was measured
using a Bruning Areograph dot planimeter.
Total surface area of the
-19prints was sufficient to insure a degree of measurement precision of
at least 90 percent.
The percentage of closed area provided objective
canopy indexes.
After the data compilation had been completed, a total of 23
physiographic factors were available for analysis.
Table 2.3 lists
the mean, median, and standard deviation for all 23 parameters.
T A B L E 2.3
SUMMARY OF PHYSIOGRAPHIC FACTORS ASSOCIATED WITH GAGE LOCATIONS
Factors
East-West Contour Orientation*
(intersections)
North-South Contour Orientation*
(intersections)
Northeast-Southwest Contour Orientation*
(intersections)
Northwest-Southeast Contour Orientation*
(intersections)
Roughness*
(intersections)
Station Elevation
(feet msl)
Average Elevation*
(feet msl)
North Average Elevation*
(feet msl)
South Average Elevation*
(feet msl)
East Average Elevation*
(feet msl)
West Average Elevation*
(feet msl)
^Computed at the four mile scale
Mean
Median
Standard
Deviation
54
46
28
67
65
31
62
63
34
55
58
25
238
240
109
5601
5060
1151
5623
5378
734
5604
5463
803
5687
5569
799
5461
5253
799
5762
5685
782
TABLE 2.3 (continued)
Factors
East-West Contour Orientation**
(intersections)
North-South Contour Orientation**
(intersections)
Northeast-Southwest Contour Orientation**
(intersections)
Northwest-Southeast Contour Orientation**
(intersections)
Roughness**
(intersections)
Average Elevation**
(feet msl)
North Average Elevation**
(feet msl)
South Average Elevation**
(feet msl)
East Average Elevation**
(feet msl)
West Average Elevation**
(feet msl)
Horizontal Canopy
(percent)
Vertical Canopy
**Computed at the one-half scale
Mean
Median
Standard
Deviation
96
60
97
107
76
95
104
69
100
108
50
107
455
265
424
5601
5152
1065
•5603
5147
1068
5614
5132
1078
5569
5082
1065
5621
5179
1090
86
89
10
6
0
11
III.
A.
DATA ANALYSIS
General Remarks
The data analysis was conducted in stages corresponding to the
three specific objectives with the determination of the relations
between meteorological parameters and precipitation receiving primary
emphasis.
Secondary considerations included I) relating physiographic
factors with precipitation variation and 2 ) evaluation of precipitation
frequency and intensity in time and space.
Stepwise multiple regres­
sion was used to study the relationships of precipitation with meteo­
rological and physiographic factors.
The third objective, to determine
the percentage of time during which the major portion of total precip­
itation fell, entailed employment of several simple sorting routines.
The Montana State University Sigma Seven computer was utilized
extensively to complete this phase of the study.
Several sorting
programs were written to arrange the data for analysis.
A stepwise
multiple regression program.prepared by the Montana State University
Mathematics Department from Biomedical Computer Programs (1967) was
employed in the major analysis.
Analysis by stepwise multiple regres­
sion was selected because independent variables as fitted in the final
equation are ranked in order of importance.
The stepwise regression program computes a sequence of multiple
linear regression equations in a stepwise manner.
variable is added to the regression equation.
At. each step one
The^variable added is
*• I--
Vtl
■23the one which makes the greatest reduction in the error sum of squares.
"Equivalently it is the variable which has the highest partial corre­
lation with the dependent variable partialed on the variables which
have already been added.
This is also the variable which, if it were
added, would have the highest F-value of the variables remaining"
(Dixon, 1967).
Variables were cycled into the equation, and if their
contribution to the equation was significant at the 5 per cent level,
they became part of the equation.
In the regression equation a variable’s F-value is equivalent to
the variance explained by the variable.
As noted above, the variable
with the largest F-value also has the highest partial correlation for
that step of the regression.
A highly significant F-value indicates
that the reduction in the variance of the dependent variable due to
an independent parameter is not likely the result of chance.
In a
regression equation those variables that reduce the variation most are
considered most significant.
Since the stepwise regression program selected only those
variables with F-values larger than a prescribed significance level,
less significant variables were not entered into the equation.
Tolerance limits were specified for 5 percent significance.
Variables entered into the equation were ranked in order of
importance, also according to F-values.
Only those factors that
explained more than 5 percent of the variance were selected as
24'
important and physically related to precipitation.
Two possible disadvantages with multiple regression analysis
include 1 ) the errors of measurement of the dependent variable,
precipitation, must be normally distributed since the multiple regres­
sion analysis assumes such a distribution and 2 ) the multiple regres­
sion equation will only develop linear relationships (Ostle, 1969),
B»
Meteorological Aspects
Initially precipitation data was stratified.
The purpose of the
stratification was to remove "noise" from precipitation data by
investigating only those precipitation events occurring under a given
set of conditions.
The given conditions were specified by the strati- -
fication design using spatial and temporal parameters.
Elevation was
chosen as a spatial stratification parameter as a result of the
physiographic analysis (See Physiographic Aspects),
Because of. the
success encountered by Yaw, _et al. (1969) with maximum surface pressure
gradient direction (MSPGD), it was selected as a temporal stratifi­
cation tool,
MSPGD is the direction from high pressure to low pressure
in which the maximum pressure difference can be found over a fixed
distance.
For this study MSPGD.was measured over a radial distance of
300 miles with the target area as a focal point.
It was calculated
from National Weather Service surface weather charts.
If no pressure
gradient existed, the surface conditions were described as high or low
pressure centers.
Analyzing frequencies of precipitation according to
'.iV:
25MSPGD suggested three groups of precipitation.
High frequencies of
precipitation were noted when the surface pressure gradient was from
of
the northeast Jjf*" from the southwest.
Also when a low pressure system
was indicated by MSPGD, frequencies of precipitation increased.
Definitions for the three MSPGD cases used in the analysis are;
Low Pressure System - no real flow direction indicated with low
surface pressure in the target area.
Northeast Flow - the maximum surface pressure gradient was from
a direction between 355° and 55°,
Southwest Flow - the maximum surface pressure gradient was from
a direction between 175° and 235°,
A separate regression analysis was performed using each MSPGD
stratification,
As noted earlier, only those factors that explained
more than 5 percent of the variance were selected as important.
The
results are summarized in Table 3,1,
For the low pressure systems cloud base temperature and 700mb
wind direction were both directly related to precipitation increases
with warm cloud base temperature and northerly to easterly winds being
associated with increased precipitation.
With warm cloud bases the
moisture capacity of the atmosphere can be expected to increase.
Increases in moisture content make increases in precipitation possible.
Precipitation increases were noted with north or northeast 700mb wind
directions.
When the airflow is from the northeast, uplift created
T A B L E 3.1
MAXIMUM SURFACE PRESSURE GRADIENT DIRECTION STRATIFICATION
Stratification
Low Pressure Systems
Multiple
Correlation
Coefficient
Factors
Sign of
Percent o1
Regression
Variance
Coefficient ■ Explained
.55
Cloud base temperature
700 mb wind direction
Northeast Flow
Southwest Flow
+
+
16.3
14.4
.
I
M
O
l
I
.42
500 mb temperature
-
17.9
Cloud base temperature
500 mb height
-
12.0
10.2
.47
-27"
by the east flanks of the Bangtails may be orographically responsible
for precipitation.
for the northeast- flow stratification, the SOOmb temperature is
the only factor explaining at least 5 percent of the variance.
cold SOOmb temperatures, precipitation increases are noted.
With
If SOOmb
temperature is related to ice nuclei formation, a physical under­
standing of its role in precipitation formation is recognized.
Cold
temperatures are conducive to increased ice nuclei concentrations.
With such increases, precipitation formation is expected and sub­
sequent increases in precipitation totals may result.
Cloud base temperature and SOOmb height appear significant for
the westerly flow case0
The cloud base temperature may again be
related to increased moisture content of the atmosphere and subsequent
increases in. precipitation.
over the area.
The low SOOmb height suggests a trough
Precipitation increases are likely results of such
conditions.
Precipitation data were also stratified according to elevation.
Consideration of station elevation suggested three elevation
divisions.
Table 3.2 lists the number of stations for each elevation
case and the extremes of elevation for each case.
’
--28-
TABLE 3.2
ELEVATION STRATIFICATION GROUPS
Stratification
Number of
Stations
Lowest
Elevation
(feet msl)
Average
Elevation
(feet msl)
Highest
Elevation
(feet msl)
14
4220
4736
5060
Medium
7
5180
5654
6200
High
6
7180
7560
7880
Low
When the regression was performed using the elevation strati­
fication, again only those factors that explained at least 5 percent
of the variance were selected as important.
The results of this
analysis are presented in Table 3.3,
For low elevation cases cloud thickness was the only important
parameter.
The direct relationship between cloud thickness and
precipitation suggests that with thick clouds more precipitation
results.
Thick clouds suggest that more moisture is available for
precipitation and also that the tops of the clouds are high, probably
implying that the cloud tops are cold.
Under such conditions ice
nuclei concentrations may increase and subsequent increases in
precipitation may result.
Cloud base temperature and cloud thickness were both directly
T A B L E 3.3
ELEVATION STRATIFICATIONS
Stratification
Low Elevation
Medium Elevation
Multiple
Correlation
Coefficient
Percent of
Variance
Explained
Cloud thickness
+
9.2
Cloud base temperature
Cloud thickness
+
+
11.2
7.6
400 millibar height
Cloud base temperature
Cloud thickness
700 millibar height
+ .
+
-
18.6
14.9
5.7
5.8
.30'
.43'
/
High.Elevation
Sign of
Regression
Coefficient
Factor
<»45
”30—
related to precipitation for the medium elevation stations.
The
physical relationship with precipitation is the same as that discussed'
for low pressure systems.
The physical understanding of thick clouds
is the same as that described for low elevation stations.
For high elevation stations, cloud base temperature and cloud
thickness were directly related with precipitation for the same reasons
as explained for low pressure systems and low elevation stations
respectively.
The 400mb and 700mb heights were both inversely related
to precipitation increases.
trough aloft.
Both factors suggest the presence of a
Under such conditions precipitation increases may result.
Seeking stronger relationships, double stratification of precip­
itation data was tried.
Separating the precipitation data into three
WSPGD groups provided one stratification.
Then dividing each WSPGD
into three elevation sections provided the second stratification.
this way a total of nine stratification groups resulted.
In
Table 3.4
summarizes the number of cases resulting from the analysis of each of
these stratification groups.
Each stratification group exhibited characteristics of a partic­
ular flow situation at a given elevation,
For example, the north-east
flow case often represented cold Arctic air moving in from the
northeast.
The effects of this airflow were studied under three
topographic divisions, those being low, medium, and high elevation
situations.
The northeast flow was investigated at each elevation to
TABLE 3.4
P R E C I P I T A T I O N C A S E S FOR E A C H S T R A T I F I C A T I O N
Stratification
Number of
Stations
( stratified )
(by elevation)
Number of
Flow cases
(stratified)
( by MSPGD )
Total ' '
Number
of
Cases
14
23
322
Low Pressure at Medium Elevation
7
23
161
Low Pressure at High Elevation
6
23
138
14
32
448
Southwest Flow at Medium Elevation
7
32
224
Southwest Flow at High Elevation
6
32
192
Northeast Flow at Low Elevation
14
29
174
Low Pressure at Low Elevation
Southwest Flow at Low Elevation
Northeast Flow at Medium Elevation .
7 .
29
203
Northeast Flow at High Elevation
6
29
174
determine characteristic relationships between precipitation and
meteorological factors.
In this way factors that were associated
with precipitation at one elevation may be distinguished from those
related to precipitation at another elevation.
This analysis
procedure was used for the southwest flow and low pressure cases also.
Due to a lack of complete radiosonde data, the stratification
operations were limited to 116 six hour intervals.
From the 116
periods, 23 low pressure cases, 29 northeast flow cases, and 32
southwest flow cases were noted in the MSPGD stratification.
The
remaining 32 periods showed random variation with respect to MSPGD.
Upon examining Table 3.5 several parameters are shown to be
important according to the percentage of variance explained for
selected stratifications.
As in the two previous analyses only those
factors explaining more than 5 percent of the variance were considered
important.
For low pressure conditions at low elevations cloud base temper­
ature was the only .important parameter.
As with the previous low
pressure system interpretation, the role of cloud base temperature is
probably related to increased moisture content of the atmosphere and
subsequent increases in precipitation.
For low pressure systems at medium elevations cloud base
temperature was again important.
The same physical interpretation
described in the initial low pressure analysis explains the importance
-33TABLE 3.5
IMPORTANT METEOROLOGICAL FACTORS
SELECTED BY STEPWISE MULTIPLE REGRESSION
Stratification
Multiple
Correlation
Coefficient
Low pressure at
low elevation gages
Factor
Sign of
Regression
Coefficient
Percent of
Variance
Explained
.44
Cloud base temperature
+
19.2
Low pressure at
.57
medium elevation gages
Cloud base temperature
'400 mb wind direction
+
26.0
6.2
Low pressure at
.71
high elevation gages
Cloud base temperature
700 mb height
+
—
28.B
21.5
Southwest flow at
low elevation gages
500 mb height
Cloud base temperature
BW
+
9.4
8.1
.42
Southwest flow at
.62
medium elevation gages
500 mb height
Cloud base temperature
+
19.5
18.4
Southwest flow at
,64
high elevation gages
Cloud base temperature
Cloud top height
+
—
24.9
18.4
Northeast flow at
low elevation gages
500 mb temperature
Cloud top height
700 mb temperature
W
.50
m
+
9.6
8.8
7.0
Northeast flow at
.61
medium elevation gages
500 mb temperature
Cloud top height
500 mb height
SB
21.2
9.4
6.8
Northeast flow at
.64
high elevation gages
500 mb temperature
*
41.0
34of cloud base temperature.
the southwest flow.
The 40Omb wind direction contribution is
West to southwest flow suggests a trough just west
of or passing through the area.
Since the major area of precipitation
in a trough is usually in the southwesterly to westerly flow, the
relationship between 4QQmb wind direction and precipitation as
suggested by the regression equation is physically realistic.
Cloud base temperature is again important for the low pressure
case at high elevations.
The same physical understanding explained
previously applies here.
The 700mb height is inversely related to
precipitation.
Such conditions suggest a trough in the area during
periods of increased precipitation.
For the southwest flow at low elevations, SOOmb height and cloud
base temperature are important.
The contribution of the SOOmb height
to the regression equation is negative, suggesting a trough aloft.
Such conditions are conducive to precipitation as suggested earlier.
The cloud base temperature is important for the same reasons noted
above.
Five hundred millibar height and cloud base temperature are shown
to be important for the southwest flow at medium elevations.
The same
physical interpretation as noted previously applies.
Cloud top height and cloud base temperature are listed as
important for the southwest.flow at high elevations.
The importance
of cloud base temperature again is explained by increased moisture
-35'
content of the atmosphere.
With cloud top height showing an inverse
relationship with precipitation, no apparent physical interpretation
is recognized and therefore, the significance of this parameter
suggests statistical chance.
The analysis of northeast flow at low elevations indicates that
SOOmb temperature, cloud top height,and 700mb temperature/are important
factors.
The importance of the 500mb temperature may be a reflection
of increased nuclei concentrations as noted with the initial northeast
flow interpretation.
There is no apparent relationship between low
cloud tops and precipitation.
Warm 700mb temperatures may reflect
increased atmospheric moisture capacity,
As with warm cloud bases,
increased moisture content may provide increased precipitation poten­
tial,
The warm lower levels indicated by 700mb temperature and the
cold upper levels designated by 500mb temperature together represent
unstable atmospheric conditions.
Such conditions increase probability
of precipitation.
Five hundred millibar temperature, cloud top height, and SOOmb
height are shown to be important for the northeast flow at medium
elevations.
The SOOmb temperature and cloud top height appear to be
important for the same reasons suggested earlier.
The positive
contribution of the SOOmb height shows no recognized interpretation
for being important and is, therefore, attributed to statistical
chance
Only the SOOmb temperature is important for the northeast flow
at high elevations.
The physical interpretation or its importance is
the same as explained in the initial northeast flow analysis.
In the analyses, several factors were related with precipitation
variation.
Among the important factors suggested by the multiple
regression equations were cloud base temperature, SOOmb temperature,
and cloud thickness.
These parameters were most frequently repeated
3
in the various regression equations.
The cloud base temperature was
the most frequent and, often, the most significant contributor to the
equations.
The direct relationship between warm cloud bases and
precipitation was attributed to increased atmospheric moisture
capacity with increased cloud base temperature.
The relationship
between cloud base temperature and precipitation is evidenced in six
of the nine stratification groups (Table 3,5),
The association between SQQmb temperature and precipitation was
noted in three of the nine stratification groups (Table 3,5),
This
relationship was attributed to increased ice nuclei concentrations
for increased precipitation formation with cold 500mb temperatures.
For this reason the importance of the 500mb temperature in relation
to precipitation as suggested by the regression equations seems
physically realistic.
Cold SOOmb temperatures should be considered
for reasons other than increased ice nuclei concentrations.
Such
temperatures suggest the actual atmospheric lapse rate may be large.
37In such cases unstable conditions are expected.
Cloud top height was shown to be important in three of the nine
stratification groups (Table 3.5).
As cloud top height was inversely
correlated with precipitation, no apparent physical interpretation for
its importance is recognized.
For this reason its significance to the
regression equation is attributed to statistical noise.
However, it
is possible that there is a mechanism involved here that is not appar­
ent at this time.
If such is the case, the statistical relationship
would be real.
Table 3.5 shows that for a given flow situation, the multiple
correlation coefficients increase with elevation.
This might be
expected since precipitation frequency increases with elevation.
With
such increases in precipitation occurrences, there are fewer zero
values.
Zero precipitation amounts tend to decrease the amount of
variation that can be explained and, hence, the multiple correlation
coefficients will decrease.
Since small precipitation totals are
noted at low elevation stations, the relative amount of variation
among these totals is less than the high elevation totals.
Again
there is less variation to explain for the low elevation stations,
and the multiple correlation coefficients may be expected to decrease.
Associating precipitation with meteorological parameters has
produced varied results.
explanations.
Linear relationships were assumed in the
The justification for these assumptions is open to
-38
question and the probability of statistical chance cannot be ignored.
Of noted importance is the role SOOtnb height exhibits with one strati­
fication and not with another.
The SOOmb height appears as both a
negative and a positive contributor to the regression equations for
different stratifications.
The physical understanding for such vari­
ation in results is not totally understood.
Some of the inconsistancy
in results can be attributed to the simple fact that .certain factors
are important only for given conditions.
Statistical accident may
explain some of the variation in results, but much of the variation
remains unexplained,
C. . Physiographic Aspects
Stepwise multiple regression was employed to find linear relation­
ships between precipitation and physiographic factors.
Of the 27
original parameters examined, four were entered in the final equation
at the 5 percent significance level.
The variables entered into the
equation are given in Table 3,6.
Elevation, highly, correlated with precipitation (.94), was listed
as the primary factor for explaining precipitation variance.
The
orographic implications of elevations best explain its importance in
the regression equation.
Orographic precipitation results from the
lifting of moist air over an elevated barrier such as a mountain range.
-39
TABLE 3.6
PHYSIOGRAPHIC INFLUENCES ON PRECIPITATION
Multiple correlation coefficient = .97
Rank
Sign of
Regression
Coefficient
Factors
F-value
Percent of
Variance
Explained
I.
Station elevation
positive
77.2
87.7
2.
Four mile roughness
positive
16.3
2.5
3.
West elevation
negative
7.6
1.7
4,
Vertical canopy
positive
5.5
1.6
Upon lifting of the moist air, adiabatic cooling results and subse­
quent condensation will occur.
formation of precipitation.
The condensate may be utilized in the
The amount designated orographic precip­
itation should not include that part of the precipitation which would
be expected from the dynamics of the associated weather disturbance,
were the disturbance over flat terrain.
Orographic precipitation is
not always limited to the ascending ground, but may extend for some
distance windward of the base of the barrier (upwind effect), and for
a short distance to the lee of the barrier (spillover).
In any case
increases in elevation suggest increases in lifting and condensation,
with subsequent increases in precipitation amounts possible.
40.
Turbulence created in the airflow may be represented as uplift
resulting from local relief.
If the roughness parameter is a
reflection of local relief, then strong and widespread turbulence
would be closely associated with extensive rough terrain under
moderate to high wind conditions.
With increased local uplift
occurring, precipitation increases may result.
The direct relation­
ship between four mile roughness and precipitation supports this
hypothesis.
The inverse relationship between west elevation and precipitation
may be a reflection of upslope-downslope characteristics.
Since air­
flow over the target area is predominantly westerly, the east flanks
of the Bangtail Mountains represent the lee side of the orographic
barrier.
If high elevation exists west of a station^ downslope flow
predominates.
With increased downslope effects precipitation is
likely to decrease.
In this way west elevation will be inversely
related to precipitation as shown in the regression equation.
The positive contribution of vertical canopy suggests increased
gage catch was due to increased shielding from winds.
In considering the results of the regression analysis, of the
23 physiographic factors used, gags elevation is the dominant factor,
with four mile roughness and west elevation making minor contributions.
D.
Precipitation Frequency with Time and Space
The term precipitation event, as used in this study, is the
■»41 ■>
recording of 0.01 inches or more of precipitation during a six hour
period at one gage.
When considering all gages over the entire season
there was an average of 128,4 precipitation events out of a possible
600 six hour time blocks.
Precipitation occurred most frequently at
the higher elevations with the maximum number of precipitation events
being 240.
The smallest recorded sum was 63 precipitation events at
the lowest elevation station.
On the average, fifty percent of the seasonal precipitation total
fell during 17.4 percent of the precipitation events.
Therefore, on
the average, half of the seasonal precipitation total came during 21.2
precipitation events.
Table 3.7 shows the relationship between
increasing percentages of the seasonal precipitation total (column 1)
and the number of precipitation events contributing to the .percentage
increase (column 2).
The percent of the total precipitation events
at each station required to achieve stated percentages of total
seasonal precipitation were averaged and are listed in Table 3.7
(column 3) as "percentage of precipitation events,"
All values listed
in Table 3.7 are seasonal averages for the 27 network stations.
The number of precipitation events contributing to the stated
percentages of the seasonal total
station.
varies, of course, from station to
This variation was analyzed by correlating the number of
precipitation events contributing to given percentages of the seasonal
total with station elevation.
This physiographic parameter was
T A B L E 3.7
VARIATION OF PRECIPITATION FREQUENCY
Percentage
of Total
Seasonal
Precipitation
Precipitation Frequency
Correlation Coefficients
Average
Number
of Events
Percentage of
Precipitation
Events
Number of
Events with
Elevation
Percentage of
Events with
Elevation
(I)
(2)
(3)
(4)
(5)
10
2.74
2,28
.84
-.49
20
5.96
5.04
.86
—.61
25
7.89
6.62
.90
-.67
40
15.11
12.52
.95
-.78
50
21.19
17.40
.96
-.78
60
28.96
23.63
.96
-.79
67
35.44
28,80
.97
— .80
75
45.48
36.71
.97
-.77
80
53.11
42.91
.97
— e76
62.70
50.20
, .97
— e74
90
75.41
60.23
.97
-.73
100
128.44
100.00
.97
85
'
43selected because of its noted contribution in the physiographic
analysis.
In Table 3,7 the .resulting correlation coefficient for
elevation is given in column 4.
Similar correlation analysis was
performed using percentages of precipitation events with elevation.
The results of this analysis are listed in column 5 of Table 3.7.
The correlation between number of precipitation events and
elevation is positive suggesting that the larger seasonal precip­
itation totals observed at higher elevations are partially due to
more frequent precipitation events.
However, the correlation between
percentage of precipitation events and elevation is negative.
This
suggests that at higher elevations a given percentage of the seasonal
precipitation total will fall during a smaller percentage of the
precipitation events, at least to the 90 percent of total seasonal
precipitation level.
Thus, those storms which make up the first 90
percent of seasonal precipitation are relatively more intense at
higher elevations when compared with lower elevations.
This is also
indicated by the fact that the correlation coefficient between the
average precipitation amounts per precipitation event and elevation
is 0,68.
Thus, more intense storms generally occur at higher
elevations.
Therefore, increased seasonal precipitation totals at
high elevations are due to both more frequent precipitation events
and to more intense precipitation during many of these events.
given percentage of the seasonal total were always equal to the
If a
-44percentage of precipitation events, then storms of uniform intensity
throughout the season would be implied.
Variation in precipitation frequency with time has shown that
increases in precipitation totals occur with increases in number of
precipitation events and greater intensity during many of those
events.
Variation of precipitation frequency with space has suggested
that a elevation is related to both frequency and intensity of precip­
itation,
As a result the increase in precipitation totals was
attributed to increasing frequency and intensity with increasing
elevation, which supports the results of the physiographic analysis.
IV.
A.
CONCLUSIONS AND RECOMMENDATIONS
Conclusions
Several implications result from this investigation.
As suggested
in the preliminary discussion, the relevance of a given parameter may
change with location and conditions; therefore, the specific contri­
bution of a given factor to the precipitation model is made with
reference to mountainous precipitation in southwestern Montana,
Five
fundamental facets of precipitation variation are concluded.
(1)
Stepwise multiple regression techniques may be used to
analyze precipitation variation.
Most factors entered into the
regression equations can be physically interpreted.
For this reason
the stepwise multiple regression analysis seems feasible for investi­
gating precipitation variation.
(2)
Cloud base temperature is positively related to variation
in precipitation.
The relationship is attributed to increased mois­
ture capacity of the atmosphere with warm temperatures.
The justifi­
cation for this interpretation is strongly evidenced by the frequent
and significant occurrence of cloud base temperature in the regression
equations,
(3)
are noted*
With decreasing SOOmb temperatures, precipitation increases
With the SOOmb temperature as an indicator of increased
ice nuclei formation, subsequent increases in precipitation are
readily interpreted.
The importance of this factor was evident for
all north-east flow cases
(4)
Variation in seasonal "precipitation exhibits a definite
control by elevation with increased totals at higher elevations.
With increasing orographic activity at higher elevations the increases
in precipitation totals are readily explained.
(5)
The increase in precipitation with elevation is a function
of increased frequency and increased intensity as indicated by
variation of precipitation events.
Cloud base temperature appears to be an important consideration
when associating upper air variables with precipitation.
In those
cases when cloud base temperature does not contribute to the regres­
sion equations, 500mb temperature has been entered into the equation.
Therefore in studying the variation of precipitation, as related to
meteorological variables, due consideration should be given to cloud
base temperature and SOOmb temperature.
Frequency and intensity of precipitation have been shown to
increase with elevation.
Such results have noted the importance of
orographic activity for explaining precipitation variation.
B.
Recommendations
Several problems arose and remained unsolved during the course of
the investigation.
In order to alleviate many of these problems, the
following recommendations for further study are suggested.
(I)
The experimental area should be removed from any precipitation
—47augmentation activities.
The experimental area for this study was
subjected to cloud seeding operations.
If the seeding activities
affected the precipitation, results should be transformed to represent
the natural situation.
If the number of no-seed cases is large enough
for valid statistical analysis, a study of just the no-seed cases
would remove any seeding effects.
If neither of these techniques are
feasible, a separate experimental area removed from any precipitation
augmentation activity is needed.
(2)
A rawinsonde program conducted in the target area to better
represent local conditions and help eliminate noise in the data is
needed.
The frequency of rawinsonde launches should be increased to
present a more detailed description of existing conditions and
launches should be made at regular intervals.
(3)
The precipitation data should be summed for both shorter
and longer intervals of time than the six hour period used.
In this
way a more detailed description of precipitation variation might be
developed.
(4 )
A study related to the determination of lag times for the
carbon hygrister should be performed.
Such information would aid
greatly in determination of cloud conditions from radiosonde records.
Information specifying lag times of radiosonde measurements is
unavailable for temperatures below DC.
(5)
Cloud determination techniques should be more refined,
A
—
48
”“
thorough investigation of radiosonde representation of clouds would
reduce problems related to cloud specifications.
(6)
An investigation with the precipitation gage network
extended over the Bridger Mountains to provide additional data needed
for a more complete understanding of the precipitation in mountainous
terrain is suggested.
The experimental area in this study was located
to the lee of the Bridger Range with prevailing westerly flow and was,
therefore, subject to rain shadow effects.
(7)
Testing of regression equations using independent precip­
itation data to help evaluate the accuracy of models should be tried.
(8)
Since noise seemed to be evident in precipitation data
before stratification, different stratification parameters might
provide more significant results*
Another possible method for
removing noise from precipitation data would be to utilize only
precipitation amounts greater than 0,10 inches in the analysis.
In
this way the errors that may be due to wind blown snow, gage errors,
etc*, should only be a small portion of the total "signal."
(9)
The investigation should be conducted on a larger sample
of data over a longer period of time,
(10)
The regression equation should be tested at individual
o
stations to determine whether or not it will serve as a prognostic
method for forecasting.
(11)
Statistical analysis of precipitation data for each
-49individual station in the network should be performed to see if the
results support the findings of the analysis on grouped data.
(12)
Different analyses should be performed with the data to
determine if any relationships not recognized by multiple regression,
exist.
By transforming the data through the use of trigonometric
functions, logrithmic functions, etc,, non-linear relationships
may be recognized.
With increasing emphasis on precipitation augmentation, the
desirability of mesoscale precipitation variation studies is enhanced.
By selecting those situations that are conducive to producing precip­
itation, one can eliminate the added expense of seeding the ineffective
cases.
In this way the maximum possible increase from seeding may be
realized.
If the regression equation proved to be a good forecasting
tool, the benefits of such investigations would be strengthened.
Mesoscale precipitation variation studies should be continued, for the
value realized from such studies could aid in the success of related
meteorological programs
APPENDIX A
DETERMINATION OF CLOUD DATA FROM RADIOSONDE OBSERVATIONS
An objective method was developed to determine the location of
cloud base height and cloud top height.
Since a cloud represents
saturated conditions, a cloud base would be selected at the point
where the relative humidity reaches 100 percent during ascent (Point
A of Figure A.I).
If in the continued ascent, the relative humidity
deviates from saturated conditions (relative humidity less than 100
percent), a cloud top is selected where saturated conditions last
exist (Point B of Figure A,I),
The implied simplification of this technique is not justified
in many cases for several reasons.
The radiosonde humidity element
(carbon impregnated hygrister) measures humidity with respect to the
dew point temperature.
The frost point temperature, analogous to the
dew point temperature, is the temperature less than 0.0C at which
atmospheric moisture will sublimate in the form of hoar frost.
The
frost point temperature is always higher than the dew point temperature
for equivalent conditions.
This temperature difference enables a
parcel of moist air to be saturated with respect to the frost point
temperature and remain unsaturated with respect to the dew point
temperature.
Figure A.2 shows that the 600mb mixing ratio with respect to the
5 1-
BOUNDARIES DETERMINED
R E LA TIVE HUMIDITY
CLOUD
TOP
HEIGHT (Millibars)
CLOUD
BY
CLOUD
BASE
25
RELATIVE
Figure A . I
50
75
HUMIDITY (%)
RELATION
DEW
BETWEEN
POINT
FROST
MIXING
POINT
AND
RATIOS
FROST
P O IN T ^ r
•40 '
TEM PERATURE
Figure A.2
(0C)
RELATIVE
MIXING
RATIO (g /k g)
HUM ID ITY (% )
^ R E L A T I V E HUMID ITY
•‘ :
v
:
-
- ■
■53dew point is always larger than the corresponding value with respect
to the frost point.
Also shown for saturated frost point conditions
is relative humidity with respect to the dew point.
From this, it is
seen that saturated conditions may exist with respect to the frost
point temperature, while the relative humidity with respect to the
dew point temperature, as measured by the radiosonde hygrister, is
considerably less than 100 percent.
Hence, a radiosonde may be in
a cloud and yet not indicate saturated conditions.
In considering cloud tops, it is found that in some instances a
trace of the humidity record shows a pronounced inflection point
where the radiosonde left the saturated or near saturated environment.
If the point of inflection is easily observed (Point B of Figure A.3),
this suggests that an abrupt change occurred in the humidity condi­
tions.
If the corresponding humidity plot indicates near saturated
conditions at point B and a' rapid decrease upward, cloud top was
selected at that height.
The subjectiveness of "near saturated conditions" rendered
necessary more objective selection criteria to define cloud boundariesg
an objective interpretation was developed.
The basis for determining
a cloud boundary was related to the minimum relative humidity at which
saturated conditions would exist.
At saturated frost point conditions
the relative humidity as measured by the carbon hygrister would be
less than 100 percent, and as the temperature decreased the deviation
CLOUD BOUNDARIES ON ORIGINAL
RADIOSONDE
HUMIDITY
CLOUD
LOUD
TOP
BASE
HUMIDITY
FREQUENCY
Figure A . 3
TRACE
- 55
from 100 percent would increase.
This expected deviation may be
justified at cold temperatures (near -400) when ice particles are
expected to exist.
However, at temperatures near -100, supercooled
cloud droplets are known to exist (Schaefer, 1946) and relative
humidity evaluated under saturated frost point conditions is not
completely valid.
A relative humidity with respect to the dew point
and calculated under saturated frost point conditions would be an
underestimate of the actual humidity situation.
Other deviations from accurate humidity measurement may result
from diurnal variations in the sensing of carbon humidity elements.
Teweles (1970) noted a possible 13 percent lower daytime relative
humidity measurement due to solar heating, deficient ventilation, long
thermal lag and internal heating within the radiosonde.
Morrissey and
Brousaides (1970) reported similar temperature induced errors in the
carbon element, with solar irradiation being the dominant factor
resulting in lower relative humidities.
Because solar irradiation
seemed to be the dominant source of error, it was surmised that the
solar effect would be reduced in a cloud environment.
Another problem encountered was the lack of information specifying
thermal lag times for the carbon hygrister at sub-freezing temperatures.
Such lag seems evident when comparing actual cloud base observations
with those derived from radiosonde traces.
On several occasions after
a radiosonde was launched, the operator at the launch site noted the
time as the instrument entered the cloud.
When comparing the time of
cloud contact by the radiosonde with the time of maximum relative
humidity measured by the radiosonde, it was noted that the maximum
relative humidity commonly was recorded about two minutes after
entrance into the cloud.
These observations were taken with cloud
base temperatures near -20C.
As noted, several sources of error confuse the actual relative
humidity situation as indicated by radiosondes.
Initially a somewhat
arbitrary value of 5 percent was selected-to represent the average
error contribution of thermal lag, solar irradiation and frost pointdew point differences.
By subtracting this 5 percent from the
relative humidity at saturated conditions with respect to the frost
point, but measured with respect to the dew point, a minimum value
for relative humidity in a cloud environment is obtained.
This
relative humidity is given by:
MRHCD =
where
f X 100%
5%
is the mixing ratio with respect to the frost point and
the mixing ratio with respect to the dew point.
Figure A.4 shows the
MRHCD values for temperatures ranging from 0 to -35C.
This particular
method was tested for cloud bases by comparing bases derived by this
method with those observed.
is
As noted above, observed values were
MINIMUM
R E LA TIV E
DETERMINATION
R E L A T IV E
HUMIDITY (%)
CLOUD
HUM IDITY
TEMPERATURE ( 0C)
Figure A.4
FOR
-
taken by radiosonde operators.
58
-
When the derived cloud bases were
subtracted from the observed, a 50mb (about -600 meters) difference,
resulted.
This technique was, therefore, abandoned for cloud base
selection.
A second cloud base determination technique was used.
The
average relative humidity at which radiosondes disappeared into
clouds as noted by the radiosonde observer was 85 percent (Super,
et al., 1971).
Since this is where the balloon disappeared, the
"base” of the cloud is somewhat lower.■ Because of this, a relative
humidity of 75 percent was chosen to be representative of cloud
base.
The objective nature of such a selection technique permits
unbiased results.
The height and temperature of the cloud base
were recorded in millibars and degrees Celsius respectively.
If the
relative humidity at the surface was greater than 75 percent, the
cloud base was selected at ground level.
Assuming that the cloud
base is at the surface under these circumstances is probably
unrealistic in some cases.
Often the inflection points of the radiosonde trace were not as
sharp or as easily defined as Point B of Figure A.3.
In such cases '
the point where the relative humidity for the trace last equaled the
MRHCD was selected as the upper cloud boundary.
In this manner the
selection of cloud top height was objective.
At times the relative humidity within a designated cloud appeared
-59unsteady (A to B of Figure A.5).
In such cases, if the relative-
humidity dropped below MRHCD during ascent, a cloud top was selected.
When the relative humidity within the cloud did not fall below the
MRHCD, continued saturated conditions were assumed.
In some cases
the MRHCD was exceeded above a point where a cloud top was selected,
indicating the presence of a second cloud layer.
In such cases, only
the lower layer was used in the analysis.
Another objective method for selecting cloud top temperature
was tried.
Chappell (1967) used the 500mb temperature as a represen­
tation for cloud top temperature in central Colorado.
When relating
this parameter with ice nuclei formation, he found it to be an
important consideration for evaluating cloud seeding operations.
Because of the elevation differences between the location of his
study and the local target area, and because the calculated cloud
top heights for the local area ranged near 600mb, the 600mb temper­
ature was selected to see if it was related to precipitation
variation.
This factor was extracted from local radiosonde data,
As noted in chapter 2, Table 2.1 lists the mean, median, and standard
deviation for each parameter extracted from the radiosonde records.
60—
UNSTEADY
RELATIVE
HUM IDITY WITHIN A CLOUD
(Millibars)
CLOUD
TOP
HEIGHT
CLOUD
BASE
RELATIVE
Figure A . 5
HUMIDITY (%)
LITERATURE CITED
Biomedical Computer Programs, ed, IijeB, Dixon,
of California Press, 1967.
Berkeley:
University
Brown, 11,3,, and E.L. Peck, "Reliability of Precipitation Measure­
ments as Related to Exposure," 3, Appl, Meteor,. 3:1, 3une,
1962.
Chappell, C.F. Cloud Seeding Opportunity Recognition. Dept, of Atm,
Sci. (Col. St. Univ,), Atmospheric Science Paper 118, 1967,
Engman, E.T., and D.M. Hershfield. Precipitation Climatology of the
Sleepers Rivers Watershed. Paper presented at the annual
meeting of the Pacific Division of AAAS, 1968,
Klein, W.H. "Precipitation Program of the Techniques Development
Laboratory," Bull, Amer, Meteor. Soc., 48:890-894, 1967.
McPartland, 3ohn T., 1971: "Snoupack Accumulation in Relation to
Terrain and Meteorological Factors in Southwestern Montana,"
Masters Thesis, Dept, of Earth Sciences, Montana State Univ,
Marlatt, William, and Herbert Riehl, 1963: "Precipitation Regimes
over the Upper Colorado River," Bourn, Geophys, Res. 68-24,
6447-6458.
Mitchell, Val L,, Arlin B. Super, and Robert H. Yaw. Atmospheric
Water Resources Program, Annual Report. Department of Earth
Sciences (Montana State University), September 1970.
Mooney, Margaret L,, and George W. Lunn. 1969: "The Area of Maximum
Effect Resulting from the Lake Almanor Randomized Cloud Seeding
Experiment," 3, Appl. Meteor., 8:68-74,
Morrissey, Barnes F., and Frederick 3. Brousaides. 1970: "Temperature
Induced Errors in the ML-476 Humidity Data," 3, Appl. Meteor.,
9:805-808.
Ostle, Bernard, Statistics in Research.
University Press, 1969.
Ames, Iowa:
The Iowa State
Packer, Paul E. 1962: "Elevation, Aspect, and Cover Effects on
Maximum Snowpack Water Content in a Western White Pine
Forest," Forest Science, Vol. 8, Number 3, 225-235.
■62
Schaefer, Vincent 3. 1946s "The Production of Ice Crystals in a
Cloud of Supercooled Water Droplets," Science, 104:457-459.'
Super, Arlin B., Robert H. Yaw, and Angelito R. Sandoval. Atmospheric
Water Resources Program, Annual Report. Department of Earth
Sciences (Montana State University), September 1968.
Super, Arlin B., Robert H. Yaw, and Angelito R. Sandoval. Atmospheric
Water Resources Program, Final Report. Department of Earth
Sciences (Montana State University), June 1969.
Super, Arlin B., Robert H, Yaw, and Val L, Mitchell. Atmospheric
Water Resources Program* Interim Progress Report. Department
of Earth Sciences (Montana State University), April 1970.
Super, Arlin B., and others. Atmospheric Water Resources Program.
Interim Progress Report, Department of Earth Sciences (Montana
State University), April 1971.
Spreen, William C., 1947: "A Determination of the Effect of Topography
Upon Precipitation," Amer. Geophys. Union. 28-2, 285-290.
Teweles1 Sidney, 1970: "Accuracy of Radiosonde Temperature and
Pressure-Height Determination," Bull. Am, Meteor. Soc., 50:386.
Yaw, R.H., A.R. Sandoval, and A.B. Super, 1968: "A Classification
System for Winter Precipitation in Southwestern Montana,"
(unpublished report). Department of Earth Sciences. (Montana
State University), 1968. (Mimeographed)
..
.-^r- iIUTVCDCTTY LIBRARIES
1762 10014346 8
.^frSe6
«378
H732
cop.2
Holman, Larry E
Some meteorological
and physiographic
aspects of winter
precipitation
IHAM« A N O A D D W K g g
Z
S c' *■/7 J*s‘-
jy ^
yC%p
iz. S rr> ,)/,
It
7
Download